News from the AI & ML world
DeeperML
Ken Yeung@Ken Yeung
//
Microsoft is aggressively pursuing AI agent technology, unveiling several key initiatives at Build 2025. Project Amelie, an experimental AI agent, is designed to build machine learning pipelines autonomously from a single prompt, effectively acting as a "mini data scientist in a box." This project leverages Microsoft Research's RD agent to automate and optimize research and development processes. Early testing shows Project Amelie outperforming current benchmarks on MLE-Bench, highlighting its potential to streamline data science workflows by automating data sourcing, analysis, and review, tasks which usually take human scientists significant amounts of time to complete.
Microsoft is also focusing on integrating AI agents across its platforms, the company announced that the GitHub Copilot coding agent has been released. The GitHub Copilot coding agent is an evolution of the Copilot, transforming it into an autonomous team member capable of handling assigned GitHub issues and creating pull requests. The agent operates asynchronously in a secure environment, analyzing code and excelling at tasks like feature addition, bug fixing, code refactoring, and documentation improvement, all while adhering to security protocols and requiring human approval for critical workflows.
Additionally, Microsoft Research has introduced Magentic-UI, an open-source, human-centered AI agent designed for collaborative web-based tasks. Unlike fully autonomous agents, Magentic-UI emphasizes transparency and user control, allowing real-time collaboration on tasks that require more than simple web searches. Built on Magentic-One and powered by AutoGen, Magentic-UI is integrated with Azure AI Foundry models and agents, enabling users to modify plans, provide feedback, and even take direct control of the browser. This focus on human-agent collaboration is a key aspect of Microsoft's broader AI strategy.
ImgSrc: thelettertwo.co
References :
Microsoft is also focusing on integrating AI agents across its platforms, the company announced that the GitHub Copilot coding agent has been released. The GitHub Copilot coding agent is an evolution of the Copilot, transforming it into an autonomous team member capable of handling assigned GitHub issues and creating pull requests. The agent operates asynchronously in a secure environment, analyzing code and excelling at tasks like feature addition, bug fixing, code refactoring, and documentation improvement, all while adhering to security protocols and requiring human approval for critical workflows.
Additionally, Microsoft Research has introduced Magentic-UI, an open-source, human-centered AI agent designed for collaborative web-based tasks. Unlike fully autonomous agents, Magentic-UI emphasizes transparency and user control, allowing real-time collaboration on tasks that require more than simple web searches. Built on Magentic-One and powered by AutoGen, Magentic-UI is integrated with Azure AI Foundry models and agents, enabling users to modify plans, provide feedback, and even take direct control of the browser. This focus on human-agent collaboration is a key aspect of Microsoft's broader AI strategy.
ImgSrc: thelettertwo.co
Share:
References :
- Ken Yeung: Microsoft’s Project Amelie Is an Experiment in ‘AI Developing AI’
- Microsoft Research: Magentic-UI, an experimental human-centered web agent
- learn.aisingapore.org: Code Agents: The Future of Agentic AI
- www.windowscentral.com: "I’ve got 30% of my time back": Microsoft Copilot reportedly helps execs quickly catch up on work after vacations